Patched bwselect

It gives different results from Matlab, at least in Octave's Image Package 2.12.0

README.md

@@ -1,7 +1,7 @@
 # MATL
 A programming language based on MATLAB/Octave and suitable for code golf.
 
-The compiler works in MATLAB R2015b or newer. Probably in older versions too, except for some specific functions. It is also compatible with Octave 4.0.0. The compiler tries to ensure consistent behaviour in both platforms. In addition, you can use it at [Try it online!](https://tio.run/#matl) and at [MATL Online](https://matl.io).
+The compiler works in MATLAB R2015b or newer. Probably in older versions too, except for some specific functions. It is also compatible with Octave 4.0.0. The compiler tries to ensure consistent behaviour in both platforms. In addition, you can use it at [Try it online!](https://tio.run/#matl) and at [MATL Online](https://matl.suever.net/).
 
 Installation: unpack the compressed file to a folder, and make that folder part of MATLAB's or Octave's search path.
 

compatibility/bwselect_comp.m

@@ -0,0 +1,15 @@
+function BW2 = bwselect_comp(BW1, C, R, N)
+% bwselect in some versions of Octave's Image Package (for example 2.12.0) gives different results
+% from those in Matlab. This fixes that. The approach used here is: find the connected components
+% and select those that contain any of the input pixels
+CC = bwconncomp(BW1, N);
+PixelIdxList = CC.PixelIdxList;
+clear CC
+lin_ind = sub2ind(size(BW1), R(:), C(:));
+BW2 = false(size(BW1));
+for c = 1:numel(PixelIdxList)
+    if any(ismember(lin_ind, PixelIdxList{c}))
+        BW2(PixelIdxList{c}) = true;
+    end
+end
+end

funDef.txt

@@ -898,7 +898,7 @@ YI	3	4	3	4	1	1	1		true	true	true	true	str = {'distinct' 'sliding'};	rearrange ma
 													if isscalar(in{3}), in{3}(2) = numel(in{1})/in{3}; end
 													if prod(in{3})~=numel(in{1}), in{3} = round(in{3}*sqrt(numel(in{1})/prod(in{3}))); end
 													out{1} = col2im(in{:});
-ZI	1	inf	2	3	1	inf	1		true	true	true	true	switch in{end}	image processing functions	Depending on numeric last input, calls an image processing function with the remaining inputs.  $0$: \matlab+imfill+. If first input is logical or numerical it is converted to char.  $1$: \matlab+bwlabeln+.  $2$: \matlab+imdilate+. This function allows second input to be number $4$, $5$, $8$ or $9$, which is interpreted as the corresponding neighbourhood mask.  $3$: \matlab+imerode+. This function allows second input to be number $4$, $5$, $8$ or $9$, which is interpreted as the corresponding neighbourhood mask.  $4$: \matlab+bweuler+.  $5$: \matlab+bwselect+ with $4$ inputs. 
+ZI	1	inf	2	3	1	inf	1		true	true	true	true	switch in{end}	image processing functions	Depending on numeric last input, calls an image processing function with the remaining inputs.  $0$: \matlab+imfill+. If first input is logical or numerical it is converted to char.  $1$: \matlab+bwlabeln+.  $2$: \matlab+imdilate+. This function allows second input to be number $4$, $5$, $8$ or $9$, which is interpreted as the corresponding neighbourhood mask.  $3$: \matlab+imerode+. This function allows second input to be number $4$, $5$, $8$ or $9$, which is interpreted as the corresponding neighbourhood mask.  $4$: \matlab+bweuler+.  $5$: \matlab+bwselect+ with $4$ inputs and $1$ output.
 													case 0
 													if islogical(in{1}) || ischar(in{1}), in{1} = double(in{1}); end
 													[out{:}] = imfill(in{1:end-1});

matl_compile.m

@@ -368,7 +368,7 @@
         'triu' 'tril' 'randsample' 'nchoosek' 'vpa' 'sum' 'mean' 'prod' 'diff' 'mod' 'repelem' 'dec2bin' 'dec2base' ...
         'hypergeom' 'disp' 'str2func' 'logical' 'circshift' 'pdist2' 'strsplit' 'max' 'min' 'strncmp' 'round'...
         'datestr' 'regexp' 'regexprep' 'imshow' 'mat2str' 'blkdiag' 'strcat' 'str2num' 'str2double' 'cconv' ...
-        'gcd' 'lcm' 'fftn' 'mode' 'nnz' 'str2sym' 'factor' 'bwlabeln' 'perms'};
+        'gcd' 'lcm' 'fftn' 'mode' 'nnz' 'str2sym' 'factor' 'bwlabeln' 'perms' 'bwselect'};
     verNumTh = [4 0 0]; % first version in which a modified function is not needed:
     if (verNum(1)<verNumTh(1)) || ((verNum(1)==verNumTh(1)) && (verNum(2)<verNumTh(2))) || ((verNum(1)==verNumTh(1)) && (verNum(2)==verNumTh(2)) && (verNum(3)<verNumTh(3)))
         fnames = [fnames {'colon'}];

spec/MATL_spec.tex

@@ -1445,6 +1445,8 @@ \subsection{MATL functions}
 This has been fixed.
 \item
 \matlab+perms+ with an input array with more than one singleton dimension gives a different result in Octave than in MATLAB (where the result corresponds to linearizing the input). This has been fixed.
+\item
+\matlab+bwselect+ in some versions of Octave's Image Package (for example 2.12.0) gives different results from those in Matlab. This has been fixed.
 \end{itemize}
 
 

spec/funDefTable/funDefTable.tex

@@ -112,7 +112,7 @@
 \matl{I} & 0 & 0-- ($^\dagger$) & paste from clipboard I \\
 \matl{XI} & 0-- (1) & 0 & copy to clipboard I \\
 \matl{YI} & 3--4 (3 / 4) & 1 & \matlab+col2im+. Uses \matlab+'distinct'+ option by default. Second and third inputs may be scalars, and then they are interpreted as numbers of columns. Third input may be a two-vector with product less then the number of elements of first input, and then it is appropriately scaled. This function allows flag strings in fourth input to be replaced by numbers, as follows: 1: \matlab+'distinct'+, 2: \matlab+'sliding'+ \\
-\matl{ZI} & 1-- (2 / 3) & 1-- (1) & Depending on numeric last input, calls an image processing function with the remaining inputs.  $0$: \matlab+imfill+. If first input is logical or numerical it is converted to char.  $1$: \matlab+bwlabeln+.  $2$: \matlab+imdilate+. This function allows second input to be number $4$, $5$, $8$ or $9$, which is interpreted as the corresponding neighbourhood mask.  $3$: \matlab+imerode+. This function allows second input to be number $4$, $5$, $8$ or $9$, which is interpreted as the corresponding neighbourhood mask.  $4$: \matlab+bweuler+.  $5$: \matlab+bwselect+ with $4$ inputs.  \\
+\matl{ZI} & 1-- (2 / 3) & 1-- (1) & Depending on numeric last input, calls an image processing function with the remaining inputs.  $0$: \matlab+imfill+. If first input is logical or numerical it is converted to char.  $1$: \matlab+bwlabeln+.  $2$: \matlab+imdilate+. This function allows second input to be number $4$, $5$, $8$ or $9$, which is interpreted as the corresponding neighbourhood mask.  $3$: \matlab+imerode+. This function allows second input to be number $4$, $5$, $8$ or $9$, which is interpreted as the corresponding neighbourhood mask.  $4$: \matlab+bweuler+.  $5$: \matlab+bwselect+ with $4$ inputs and $1$ output. \\
 \matl{J} & 0 & 0-- ($^\dagger$) & paste from clipboard J \\
 \matl{XJ} & 0-- (1) & 0 & copy to clipboard J \\
 \matl{ZJ} & 1-- (2 / 3) & 0-- (1) & Depending on numeric last input, calls a symbolic-specific function with the remaining inputs.  $0$: \matlab+simplify+.  $1$: \matlab+pretty+.  $2$: \matlab+latex+.  $3$: \matlab+numden+.  \sa \matl{X\$} \\